SimpleSense Logo

Trial Layout

Layout Summary

include product application date (if applicable) Note: Need to clean up IO file organization in the ‘Report’ folder

Cropping System Detail
Parameter Value
Crop Corn
Cultivar Beck 6365AM
Plant Date 2024-04-18
Seed Rate 31000
Application Date 2025-03-23
Acres
ft²
Length (ft)
Width (ft)
Study Area 1.15 50123.53 905.34 55.36
Plot Size (Avg) 0.02 817.29 151.04 5.41

Experimental Design: Randomized Complete Block Factorial.

Type Count
Plots 48
Replications 4
Treatments 12

NOTE: Need to integrate a new and more evolved data set representing realistic treatments and then actually use the document to tell the story.

Summary of Treatment Variables
Treatment Factor A Factor B
A Product A Time 1
B Product A Time 2
C Product A Time 3
D Product A Time 4
E Product B Time 1
F Product B Time 2
G Product B Time 3
H Product B Time 4
I Product C Time 1
J Product C Time 2
K Product C Time 4
K Product C Time 3

Data Aquisition

Equipment

DJI MAVIC 3M

DJI Mavic 3M

Multispectral Imaging System + RGB

Highly integrated imaging system

Imaging system has one 20MP RGB camera and four 5MP multispectral cameras (green, red, red edge, and near infrared), enabling applications such as high-precision aerial surveying, crop growth monitoring, and natural resource surveys.

5MP multispectral camera

  • Near-infrared (NIR) 860 nm ± 26 nm
  • Red edge (RE) 730 nm ± 16 nm
  • Red (R) 650 nm ± 16 nm
  • Green (G) 560 nm ± 16 nm

RGB Camera Features

  • 4/3 CMOS - 20MP Image Sensor
  • 1/2000s - Fastest mechanical shutter speed
  • 0.7s - High speed burst when using RGB camera

Sunlight sensor

The built-in sunlight sensor captures solar radiation and records it in an image file, allowing light compensation of the image data during 2D reconstruction. This results in more accurate NDVI results as well as increased accuracy and consistency of the acquired data over time.

RTK module

Precise images that capture every pixel. Mavic 3M with RTK module for centimeter-level positioning. The flight controller, camera, and RTK module synchronize within microseconds to accurately capture the location of each camera’s imaging center. This allows Mavic 3M to perform high-precision aerial surveys without using ground control points.

Flight Parameters

Parameter Summary
Parameter Value
Image Count 229
Height Above Ground Level (m) 76.2
Ground Sampling Distance (m) 0.05
Focal Length (mm) 0.672
Width x Height (px) 2048 x 1536

Make table subsections for internal v external parameters. Need to include IFOV?

Data Capture

More specific flight parameter information on the ODM Quality Report.

Time of Aquisition
Date Start End Duration
06/23/21 12:43:40 12:53:33 9 min, 53 sec

Ground Control

Ground Control Points (GCPs) are crucial for agriculture drone mapping, especially for research purposes. They ensure the precise georeferencing of aerial images, correcting any distortions and aligning the maps with real-world coordinates. By using GCPs, researchers can obtain reliable and actionable data, enabling them to make informed decisions and improve agricultural practices.

EPSG: 4326
Ground Control Points
id latitude longitude
1 40.31178 -85.14850
2 40.31142 -85.14803
3 40.31178 -85.14791
4 40.31142 -85.14841
5 40.31187 -85.14820

Weather Conditions

Cloud Descriptions

Cloud Descriptions

Sun Angle

Growing Degree Day (GDD)

The formula for calculating Growing Degree Days (GDD) is as follows:

\[ \text{GDD} = \frac{T_\text{max} + T_\text{min}}{2} - T_\text{base} \]

where:

\[ T_\text{max} \text{ is the daily maximum temperature,} \]

\[ T_\text{min} \text{ is the daily minimum temperature,} \]

\[ T_\text{base} \text{ is the base temperature below which growth is negligible.} \]

Requirements:

Integration of temperature data derived from the most local weather station.

  • Stream temp data
  • Base temperature
  • Date/ time of measurement
## [1] "Wind speed, temperature, cloud cover & sun angle during the time imagery was collected. Would like to find a way to get weather data dynamically by using location and aquisition time. This should be considered across multiple flight times - one report or a summary report?"
## $example
## # A tibble: 5 × 5
##   id          name                     latitude longitude distance
##   <chr>       <chr>                       <dbl>     <dbl>    <dbl>
## 1 USC00129427 WEST LAFAYETTE PURDUE        40.4     -86.9    0.867
## 2 US1INTP0005 WEST LAFAYETTE 1.6 S         40.4     -86.9    0.938
## 3 USC00129435 W LAFAYETTE SEW PLT          40.4     -86.9    1.51 
## 4 US1INTP0074 WEST LAFAYETTE 0.9 SE        40.4     -86.9    2.00 
## 5 USW00014835 LAFAYETTE PURDUE UNIV AP     40.4     -86.9    2.57
## # A tibble: 1 × 1
##   id         
##   <chr>      
## 1 USC00129427
## # A tibble: 0 × 2
## # ℹ 2 variables: id <chr>, date <date>
##         date tmin tmax mean_temp gdd
## 1 2025-02-01    5   15        10   0
## 2 2025-02-02    7   17        12   2
## 3 2025-02-03    8   18        13   3
## 4 2025-02-04    6   16        11   1
## 5 2025-02-05    9   19        14   4
## 6 2025-02-06   10   20        15   5
## 7 2025-02-07   11   21        16   6

Image Processing

Workflow

Simplified overview of the processing steps

Specify the software and tools used for processing the UAV imagery.

Processing Steps in WebODM

In our small plot research, we use WebODM to process multispectral imagery captured by UAVs. This ensures we can accurately compare treatments and gather valuable insights into crop performance. Here are the key steps involved:

  1. Image Calibration:
    • We start by calibrating the multispectral images to ensure accurate color representation and reflectance values.
    • Calibration panels captured during the UAV flight are used to adjust the raw images for consistent lighting conditions.
  2. Ground Control Points (GCPs):
    • Ground control points are incorporated to improve the spatial accuracy of the processed images.
    • These GCPs are markers placed in known locations within the field, surveyed with precise GPS coordinates.
  3. Orthomosaicking:
    • The calibrated images are stitched together to create a single, seamless high-resolution orthomosaic.
    • This involves aligning and blending overlapping images to produce a detailed map of the entire field, correcting for any distortions.
  4. Digital Surface Model (DSM) Generation:
    • Along with the orthomosaic, we generate a digital surface model that represents the elevation of the field’s surface.
    • This helps in understanding the topography and detecting any variations in elevation that might affect crop growth.
  5. Quality Assurance:
    • A thorough quality check is conducted to ensure the accuracy and reliability of the processed outputs.
    • This includes validating the orthomosaic, DSM, and vegetation indices against the ground truth data and making necessary adjustments.

Further information available on the ODM Quality Report.

Processing steps in SimpleSense

  1. Vegetation Indice Maps:
    • Image bands are manipulated to exploit features for segmentation or determine effect of imposed treatments.
  2. Plot Layer Creation:
    • Plots are built in a GIS and integrated with customers experimental design.
  3. Statistical Analysis:
    • Zonal statistics for imposed treatments.
    • Analysis of Variance (ANOVA) to test if there are significant differences in NDVI values between different treatment groups.
    • Post-hoc Analysis with Tukey’s HSD if ANOVA indicates significant differences, we use Tukey’s HSD test to identify which specific groups differ from each other.
  4. Visualizations & Findings:
    • Figures and tables that are ready to be integrated
  5. Outlier Detection:
    • In-season directive to potential outlier plots.
  6. Organized Deliverables:
    • Report
    • Spatial data layers
    • Tabular data layers
    • Figures and tables
    • Statistical outputs
    • Imagery (if desired)

Indices Calculation

NDVI

The Normalized Difference Vegetation Index (NDVI) is a measure of vegetation health and density. For the Mavic 3M, the specific wavelengths for the bands are:

  • RED band: 650 nm

  • NIR band: 860 nm

The NDVI formula with these specific wavelengths is:

\[ \text{NDVI} = \frac{(\text{NIR}_{860} - \text{RED}_{650})}{(\text{NIR}_{860} + \text{RED}_{650})} \]

Where:

  • NIR\(_{860}\) represents the near-infrared light (860 nm) reflected by vegetation.

  • RED\(_{650}\) represents the red light (650 nm) reflected by vegetation.

NDRE

The Normalized Difference Red Edge Index (NDRE) is a measure of vegetation health and density, similar to NDVI, but it uses the red-edge band instead of the red band. It is calculated using the formula:

\[ \text{NDRE} = \frac{(\text{NIR} - \text{RE})}{(\text{NIR} + \text{RE})} \]

Where:

  • NIR represents the near-infrared light reflected by vegetation.

  • RE represents the red-edge light reflected by vegetation.

VARI

The Visible Atmospherically Resistant Index (VARI) is a measure of vegetation health that is less sensitive to atmospheric effects. It is calculated using the formula:

\[ \text{VARI} = \frac{(\text{GREEN} - \text{RED})}{(\text{GREEN} + \text{RED} - \text{BLUE})} \]

Where:

  • GREEN represents the green light reflected by vegetation.

  • RED represents the red light reflected by vegetation.

  • BLUE represents the blue light reflected by vegetation.

Statistical Analysis

Note: Probably should consider moving the statistic summary to this section? Should also consider bringing the GDD to this section

Randomized Complete Block Factorial

Logic

We use ANOVA (Analysis of Variance) and Tukey’s HSD (Honestly Significant Difference) test to analyze remote sensing vegetative reflectance data collected from agricultural research plots. ANOVA helps to test if there are significant differences in the mean NDVI values between different treatment groups, and Tukey’s HSD test identifies which specific groups differ.

ANOVA

ANOVA helps to test if there are significant differences in the mean vegetative reflectance values between different treatment groups. The formula for ANOVA is:

\[ F = \frac{\text{Between-group variability (Mean Square Between)}}{\text{Within-group variability (Mean Square Within)}} \]

where:

\[ \text{Mean Square Between} = \frac{\text{Sum of Squares Between}}{\text{Degrees of Freedom Between}} \]

\[ \text{Mean Square Within} = \frac{\text{Sum of Squares Within}}{\text{Degrees of Freedom Within}} \]

Tukey’s HSD Test

If ANOVA indicates significant differences, we use Tukey’s HSD test to identify which specific groups differ. The formula for Tukey’s HSD is:

\[ \text{HSD} = q_{\alpha} \sqrt{\frac{\text{Mean Square Within}}{n}} \]

where \(q_{\alpha}\) is the studentized range statistic and \(n\) is the sample size of each group.

Interpretation

  1. ANOVA Results: Examine the ANOVA results to determine if there are significant differences between treatment groups based on the F-statistic and p-value.

  2. Tukey’s HSD Results: Use Tukey’s HSD test results to identify which specific treatment groups have significantly different NDVI values.

Inference across multiple NDVI measures

Growth Stage - V4

Summary Statistics and Tukey’s HSD Test for VARI_mean by Treatment
Treatment Mean Median Min Max N NDVI_M_1 groups
A 0.5845913 0.5845913 0.5845913 0.5845913 4 0.5845913 a
B 0.4270198 0.4270198 0.4270198 0.4270198 4 0.4270198 f
C 0.5260286 0.5260286 0.5260286 0.5260286 4 0.5260286 b
D 0.3550845 0.3550845 0.3550845 0.3550845 4 0.3550845 g
E 0.4939980 0.4939980 0.4939980 0.4939980 4 0.4939980 d
F 0.5028109 0.5028109 0.5028109 0.5028109 4 0.5028109 c
G 0.3514918 0.3514918 0.3514918 0.3514918 4 0.3514918 g
H 0.3104568 0.3104568 0.3104568 0.3104568 4 0.3104568 h
I 0.5791387 0.5791387 0.5791387 0.5791387 4 0.5791387 a
J 0.4222403 0.4222403 0.4222403 0.4222403 4 0.4222403 f
K 0.4743898 0.4707931 0.4707931 0.4923733 4 0.4761882 e
L 0.4923733 0.4923733 0.4923733 0.4923733 4 0.4923733 d

Growth Stage - V9

Summary Statistics and Tukey’s HSD Test for NDVI by Treatment
Treatment Mean Median Min Max N NDVI_M_2 groups
A 0.7427292 0.7427292 0.7427292 0.7427292 4 0.7427292 a
B 0.5059123 0.5059123 0.5059123 0.5059123 4 0.5059123 f
C 0.5098803 0.5098803 0.5098803 0.5098803 4 0.5098803 ef
D 0.4294196 0.4294196 0.4294196 0.4294196 4 0.4294196 g
E 0.5488135 0.5488135 0.5488135 0.5488135 4 0.5488135 de
F 0.6144698 0.6144698 0.6144698 0.6144698 4 0.6144698 bc
G 0.6222045 0.6222045 0.6222045 0.6222045 4 0.6222045 b
H 0.5294373 0.5294373 0.5294373 0.5294373 4 0.5294373 ef
I 0.5779646 0.5779646 0.5779646 0.5779646 4 0.5779646 cd
J 0.5404358 0.5404358 0.5404358 0.5404358 4 0.5404358 def
K 0.6194858 0.6385688 0.5240707 0.6385688 4 0.6099443 bc
L 0.5240707 0.5240707 0.5240707 0.5240707 4 0.5240707 ef

Growth Stage - R1

Summary Statistics and Tukey’s HSD Test for NDVI by Treatment
Treatment Mean Median Min Max N NDVI_M_3 groups
A 0.6649717 0.6649717 0.6649717 0.6649717 4 0.6649717 d
B 0.8124048 0.8124048 0.8124048 0.8124048 4 0.8124048 a
C 0.6751085 0.6751085 0.6751085 0.6751085 4 0.6751085 d
D 0.6969188 0.6969188 0.6969188 0.6969188 4 0.6969188 cd
E 0.4772124 0.4772124 0.4772124 0.4772124 4 0.4772124 e
F 0.6599751 0.6599751 0.6599751 0.6599751 4 0.6599751 d
G 0.7379817 0.7379817 0.7379817 0.7379817 4 0.7379817 bc
H 0.7389898 0.7389898 0.7389898 0.7389898 4 0.7389898 bc
I 0.7045758 0.7045758 0.7045758 0.7045758 4 0.7045758 cd
J 0.4961169 0.4961169 0.4961169 0.4961169 4 0.4961169 e
K 0.8008585 0.8287327 0.6614873 0.8287327 4 0.7869214 ab
L 0.6614873 0.6614873 0.6614873 0.6614873 4 0.6614873 d

Growth Stage - R3

Summary Statistics and Tukey’s HSD Test for NDVI by Treatment
Treatment Mean Median Min Max N NDVI_M_4 groups
A 0.8481187 0.8481187 0.8481187 0.8481187 4 0.8481187 c
B 0.7322143 0.7322143 0.7322143 0.7322143 4 0.7322143 e
C 0.8908149 0.8908149 0.8908149 0.8908149 4 0.8908149 b
D 0.7555652 0.7555652 0.7555652 0.7555652 4 0.7555652 d
E 0.6943518 0.6943518 0.6943518 0.6943518 4 0.6943518 f
F 0.8421285 0.8421285 0.8421285 0.8421285 4 0.8421285 c
G 0.8767229 0.8767229 0.8767229 0.8767229 4 0.8767229 b
H 0.6039105 0.6039105 0.6039105 0.6039105 4 0.6039105 g
I 0.9174953 0.9174953 0.9174953 0.9174953 4 0.9174953 a
J 0.4638537 0.4638537 0.4638537 0.4638537 4 0.4638537 h
K 0.8300481 0.8210227 0.8210227 0.8751748 4 0.8345607 c
L 0.8751748 0.8751748 0.8751748 0.8751748 4 0.8751748 b

Inference across multiple NDVI measures

Visualization and Reporting

Notes: - Could include maps of all 6 timings, and then compute statistics on these values? Should they be interactive? TMI for interactivity?
Could it be something of a GIF with changing figure? Should have only one legend to represent all of the figures - Need to display multiple VI - for example NDVI and NDRE

– Need to standardize the legend labels for all of the different growth stages.

Growth Stage - V4

Growth Stage - V9

Growth Stage - R1

Growth Stage - R3

Findings

References

Thank you for your support!

Richard M. Smith, CCA, Ph.D

SimpleSense Logo